Enamel Structure in Astrapotheres and Its Functional Implications

Home , Astrapotheria, Astrapotherium, Scaglia (mammal)

Scanning Microscopy

Volume 6 Number 2 Article 15

6-24-1992

John M. Rensberger University of Washington

Hans Ulrich Pfretzschner University of Bonn

Follow this and additional works at: https://digitalcommons.usu.edu/microscopy

Part of the Biology Commons

Recommended Citation Rensberger, John M. and Pfretzschner, Hans Ulrich (1992) "Enamel Structure in Astrapotheres and Its Functional Implications," Scanning Microscopy: Vol. 6 : No. 2 , Article 15. Available at: https://digitalcommons.usu.edu/microscopy/vol6/iss2/15

This Article is brought to you for free and open access by the Western Dairy Center at DigitalCommons@USU. It has been accepted for inclusion in Scanning Microscopy by an authorized administrator of DigitalCommons@USU. For more information, please contact [email protected]. Scanning Microscopy, Vol. 6, No. 2, 1992 (Pages 495-510) 0891- 7035/92$5. 00 +. 00 Scanning Microscopy International, Chicago (AMF O'Hare), IL 60666 USA

ENAMEL STRUCTURE IN ASTRAPOTHERES AND ITS FUNCTIONAL IMPLICATIONS

John M. Rensberger* 1 and Hans Ulrich Pfretzschner 2

1Department of Geological Sciences and Burke Memorial Washington State Museum, University of Washington, Seattle; 2Institut fiir Palaontologie, University of Bonn, Bonn, Germany.

(Received for publication February 17, 1992, and in revised form June 24, 1992)

Abstract Introduction

Astrapotheres, large extinct ungulates of South Mammalian teeth have a great range of form and these America, share with rhinoceroses vertical prism decussation differences have been used extensively to determine phyletic in the cheek tooth enamel. The similarity extends beyond relationships and evolutionary histories. The dental diversity in merely the direction of the planes of decussation. The mammals derives in part from the diversity of dental function vertical decussation in astrapotheres is confined to the inner and behavioral use of teeth, which include food gathering and part of the enamel and has uniformly well-defined zones in processing, offensive and defensive behaviors as well as social which the prism direction differs by nearly 90° and the zones activities (Crompton & Hiiemae, 1969). are separated by narrow transitional borders of intermediate It is well known that the microstructure of mammalian prism direction. The outer enamel consists of predominantly tooth enamel varies considerably among taxa (e.g. occlusally and outwardly directed prisms. Within the outer Korvenkontio, 1934, Kawai, 1955; Boyde, 1978). Recently, enamel is a region of horizontally decussating prisms; here evidence has been found that differences in enamel the angle of decussation is usually smaller than that of the microstructure in at least some taxa have not arisen randomly inner vertically decussating prisms. but somehow represent responses to selective factors in the Except for the horizontal decussation in the outer chewing mechanics and dietary adaptations. A directional enamel, these conditions match structures that have been asymmetry of the enamel microstructure within single molars described for rhinocerotoids. These features, together with of arvicoline rodents is related to the direction of motion of the the similarity in premortem crack direction and gross shape jaws during mastication (von Koenigswald, 1980). A of the cheek teeth, imply that astrapotheres and reorientation of the microstructural fabric in tapiroids and rhinocerotoids shared essentially the same system of cheek rhinocerotoids, in which the occlusal surface has become tooth mechanics. highly lophodont, is related to the attitude of the functional However, the microstructure of the canine enamel in edges of the lophs and hence to the masticatory mechanism the astrapotheres is distinct. The lower canine enamel of the (Rensberger & von Koenigswald, 1980; Fortelius, 1985; Oligocene Parastrapotherium exhibits a form of vertical Boyde and Fortelius, 1986). Prism decussation is weak or decussation modified by a wavelike bending of prism zones, absent in the soft food eating ceboid primate Alouatta but well whereas the decussation in the rhinocerotoid canine is developed in Cebus apella, C. albifrons, and Chiropotes, horizontal. The lower canine in Parastrapotherium was whose diets include hard objects (Maas, 1986). The directions subjected to different loading conditions, judging from of the prisms in the molars of the koala and in the opossum multiple sets of premortem crack directions. The modified are related to the direction of the occlusal force (Young, et al., vertical decussation would in theory resist cracking under 1987; Stem, et al.,1989). different directions of tensile stresses. This is confirmed by The ubiquity of enamel in the teeth of vertebrates from the sinuous paths of cracks that run in directions differing by fishes to mammals, its hardness, and its arrangement as a thin up to 90°. That diverse stresses were generated in the enamel outer layer together leave little doubt that its fundamental during life is confirmed by the pattern of premortem cracks in function is to resist abrasive wear. Enamel is anisotropic in its Parastrapotherium. The enamel in the upper canine of a late response to abrasive wear (Rensberger and von Koenigswald, Miocene astrapothere lacks decussation but may have 1980) which derives from the fact that apatite crystals abrade resisted cracking under varied loading conditions by virtue of less quickly in a direction parallel to the C axis than in a a 3-dimensional wavelike bending of the prisms. direction normal to the C axis. In the outer enamel, the C axes of the crystallites tend to be directed toward the outer surface, Key Words: Enamel microstructure, molar, canine, which minimizes the rate of wear in a direction normal to the stresses, crack patterns, fracture resistance, functional morph surface. ology, convergent evolution, Astrapotheria, Astrapotherium, The major structural limitation of materials with Parastrapotherium, Rhinocerotoidea. ceramic-like properties is their brittleness (Clegg, et al., 1990). The brittleness of enamel makes it subject to cracking and thus * Address for correspondence: to initiating failure of the tooth. The tendency to crack limits Department of Geological Sciences the degree of stress concentration that can be developed in and Burke Memorial Washington State Museum teeth, which in tum limits the toughness and hardness of food University of Washington DB-10 materials that can be physically reduced to small particles by the Seattle, Washington 98195 masticatory system and the speed with which large volumes Phone No.: (206) 543-7036; FAX: (206) 543-9285 can be processed. A conspicuous correlate of the early

495 J. M. Rensberger & H. U. Pfretzschner

Cenozoic diversification of mammals was the increase in body the enamel microstructure of astrapothere cheek teeth with that size, which raised the maximum level of stress attainable at the in the rhinocerotoids, describe the microstructure in occlusal surface through stress concentrating mechanisms. astrapothere canine enamel, which differs from that in the Correlating with larger body sizes was the presence of a new cheek teeth, and consider the functional implications of these microstructure that improved the crack resistance of enamel conditions for tooth use. (von Koenigswald, et al., 1987) and allowed higher stresses to Institutional Abbreviations be attained before fracture. High stress concentrations FMNH: Field Museum of Natural History, Chicago, resulting in tooth loss occur as an accidental and probably Illinois. frequent event in lower vertebrates, but the evolutionary GPIBO: Institut fiir Palaontologie der Universitat, solution of more rapid tooth replacement was not available to Bonn, Germany diphyodont mammals, in which the masticatory system is UCMP: University of California Museum of based on precise occlusion (Hopson & Crompton, 1969; Paleontology, Berkeley, California. Crompton & Sun, 1985). UWBM: Burke Memorial Washington State Museum, Thus, owing to high stresses developed in their teeth and University of Washington, Seattle, the diphyodont tooth replacement scheme, enamel in Cenozoic Washington mammals of medium to large size must have been subject to selection for both crack resistance and abrasion resistance. Methods Fracturing is still a limiting factor in modern Carnivora that heavily load their teeth in the course of eating bones (Van The scanning electron microscope (SEM) examinations Valkenburgh & Ruff, 1987; Van Valkenburgh, 1988). of astrapothere enamel microstructure are based on tooth Because both enamel and dentin are denser than other fragments of Parastrapotherium (Deseadan, early Oligocene of hard tissues, teeth are the best preserved structures in the Pico Truncado, Santa Cruz, Argentina), Astrapotherium sensu ma."Illilalian fossil record. Because of this large resource and lato (Friasian, late Miocene La Venta Fauna, Colombia, S.A.), its potential information about dietary behaviors, a number of and Astrapotherium sp. (Santacrucian, early Miocene, workers have attempted to interpret fossil diets by investigating Argentina) that are too incomplete to provide traditional the relationships between striae, pits, and other features of quantitative data about gross tooth morphology and systematic abrasive wear preserved on both fossil and modern tooth relationships. Similar fragments of Subhyracodon (Chadron enamel (Teaford, 1988, reviews this literature). As our Formation, ~arly Oligocene, South Dakota), a rhinocerotoid, understanding of how food and teeth interact in processing were examined for comparison, along with published food (Lucas & Luke, 1984; Janis & Fortelius, 1988), how descriptions of rhinocerotoid enamel. For the canine enamel enamel microstructure influences wear patterns (Maas, 1991), structure, we removed an inconsequential fragment from the and how enamel microstructural differences are related to margin of an existing fracture in the canine of two taxa. Each differences in the occlusal mechanics of teeth becomes more prepared enamel fragment is identified by a suffix added to the complete, our ability to utilize this large potential source of original museum catalogue number and is being returned to the information about the evolution of feeding behaviors will original museum collection along with a description of the proportionately improve. position and orientation of the fragment on the tooth. One of the most widespread types of mammalian enamel Secti_ons of enamel were examined in three planes (Fig. consists of Hunter-Schreger bands (HSB), which are layers of 1): tangential (parallel to the outer surface), radial (parallel to alternately directed prisms. This was the enamel type that the long axis of the tooth and normal to the outer surface) and appeared in at least a few mammals in the early Paleocene and transverse (normal to the long axis of the tooth). Figured initiated the special strengthening of enamel that accompanied tangential and radial sections are oriented with the occlusal the mammalian diversification (von Koenigswald et al., 1987). direction toward the top of the figure unless otherwise indicated Prisms fracture less easily perpendicular than parallel to the by a hollow arrow. long prism axes (Boyde, 1976; Rasmussen et al., 1976) and Because of the anisotropic behavior of enamel to the strength enamel gains through reinforcement by HSB fracturing as well as to corrosion by acid, the microsrructure is derives from the way crack paths are forced to change direction visible in either broken surfaces or polished and acid-etched by decussating prism directions (Pfretzschner, 1988). surfaces. All of the astrapothere teeth in the FMNH and The primitive condition for the HSB is one in which the U~MP collections were examined under low power light planes between the layers of alternating prism directions are microscopy. Most of the transverse and radial sections of more or less horizontal, as viewed tangentially. At the other enamel fragments were first examined as broken, unpolished extreme of HSB attitude is the structure in rhinocerotoids, in and unetched surfaces with the SEM. The relief of broken which a 90° rotation of the Hunter-Schreger planes has sections gives an indication of the resistance of the occurred and the HSB and decussation planes are vertical. In microstructure to fracturing and allows recording structural vertical decussation, the adjacent groups of prisms intercept a differences at low SEM magnifications of large areas of transverse (horizontal) plane at different angles. In horizontal enamel. These sections were subsequently examined as decussation, the adjacent groups of decussating prisms are nearly parallel to a transverse plane but intercept a radial metacone paracone (vertical) plane at different angles. + I metastyle radial Fortelius (1985: Figs. 12, 38, 39) reported vertical \ section tram verse transverse decussation in the cheek tooth enamel of sp., a Astrapotherium section ~h I Miocene member of the Astrapotheria, a group of large mammals that were confined to South America until their radial~n extinction late in the Cenozoic. Because vertical section ~ tangential Hunter-Schreger bands are rare in mammals, and this condition ~tangentia section almost certainly arose independently in the two lineages, the section degree of similarity of the structure in astrapotheres to that of Upper Cheek Tooth (labial view) Canine rhinocerotoids is of particular functional interest because any similarities must have resulted from similar selective pressures rather than from a shared ancestry. In this study we compare Figure I. Orientation of sections.

496 Enamel Structure & Function in Astrapotheres

polished -;ections etched in dilute hydrochloric acid for 5 to 15 Miocene, and reflected light microscopic examination of molar secondls. The polished sections were either initially ground enamel sections and surfaces in Parastrapotherium. using couse (150 or 360) grit silicon carbide wet-dry paper Vertical decussation followed by finer paper through 1200 grit, or ground with a Sets of prisms running in different directions are most silicon ca-bide stone followed by polishing with a wet slurry of clearly defined in sections normal to the decussation plane. In silicon carbide powder. For tangential sections, the outer the case of vertical decussation, transverse and tangential enamel was removed into the zone of vertical decussation, sections offer good zone delineation. The structure in each of which va;es in depth, either by air abrasion with dry aluminum these views is discussed below. oxide povder or by grinding. In transverse sections of molar enamel in astrapotheres A :heoretical prediction of the loading directions of the (Fig. 2), the alternating HSB are as prominent and uniformly canine in life was obtained by calculating the principal stresses defined as in rhinocerotoids. The prisms of one of these zones for a finite element model, using the microcomputer application correspond in structure to the set labelled B in rhinocerotoids MSC/pall, and comparing the cracks predicted by the model to by Rensberger and von Koenigswald (1980) and begin those obs::rved in the canine enamel. growing in an obliquely outward and cervical direction from Empirical data for stress directions were found by the enamel-dentin junction (EDJ). Then, as the zone narrows identifying premortem cracks in the enamel. Cracks were slightly near the middle of the enamel thickness, the prisms of recogniz~d as premortem when their edges bear a polish this set become horizontal (Fig. 3). The prisms of the adjacent resulting from chewing or other usage of the teeth (Rensberger, set on either side (corresponding to set A in rhinocerotoids) run 1987). The evidence that such abrasion is due to premortem occlusally and outward. This set transversely widens slightly activities and not to postmortem depositional processes is the in the middle of the enamel and appears to continue with little differenti1l distribution of the polish, especially its confinement change toward the outer edge (Fig. 2). Higher magnification is near the occlusal ·surface of the cheek teeth or near the tip of the needed to show the reversed directions of the prisms of sets A canine. and B (Fig. 4). Sets A and B are separated by a thin band of Cr.1ck directions in the canines were measured by transitional prisms of almost horizontal attitude (Fig. 4). photogra hing and digitizing the cracks and analyzing the In radial sections of astrapothere enamel, the inner directions with a microcomputer application, Macazimuth. enamel shows the strong angular divergence of prisms sets A This application, in effect, repeatedly steps one pixel at a time and B (Fig. 5) that characterizes rhinocerotoid enamel along the graphic line representing a crack and measures the (Rensberger and von Koenigswald, 1980, Figs. 7, 8). In the direction of a point on the line at a specified distance ahead. outer part of the enamel the distinctiveness of these zones As the algorithm proceeds, the part of the line covered is disappears and all prisms are outwardly and upwardly directed recorded separately so that no lines are retraced as a result of (Fig. 5). To achieve this occlusal-outward slant, the prisms crack junctions. The program is written in FORTH and runs with an initial cervical-outward direction (set B) bend abruptly on a Macintosh computer. in the middle region of the enamel (Fig. 6). In mis study, each azimuth is the angle in a plane tangent The astrapotheres strongly resemble the rhinocerotoids in to the enamel surface measured clockwise from the lingual all these details (Rensberger and von Koenigswald, 1980, direction to the crack direction (the direction parallel to the Figs. 3, 4, 5, 7, 8); Boyde and Fortelius, 1986, Figs. 20, longitudinal or cervical-occlusal axis of the tooth would be 90°. 21-23). The direction of the crack is based on the direction formed by a As a consequence of the combination of sharply line connecting a given pixel on the crack with the next pixel differentiated zones of vertically decussating prisms and the whose x or y coordinate is exactly 6 pixels distant. This allows equally distinct thin zone of transitional prisms with horizontal discrimination of 24 directions, an increase in precision attitude, the more or less horizontal occlusal surfaces of molar compared to a similar algorithm used by Rensberger, Forsten enamel bear a ridge and valley relief (Fortelius, 1985, Fig. 38) and Fortelius (1984). In the latter algorithm, each azimuth was like that in rhinocerotoids (Rensberger and von Koenigswald, based on the second closest pixel ahead, yielding 8 possible 1980, Fig. 2; Fortelius, 1985, Figs. 18, 32, 38). The relief is directions. For additional details about the method, see caused by the dependence of wear resistance of hydroxyapatite Rensberger, et al. (1984). crystals upon their alignment and the tendency for the average The 24 azimuths discriminated by 6 pixel separation are: direction of crystallites making up a prism to be parallel to the 0°, 9.5°, 18.4°, 26.6°, 33.7°, 39.8°, 45°, so.2°, 56.3°, 63.4°, long axis of the prism. 71.6°, 80.5°, 90°, 99.5°, 108.4°, 116.6°, 123.7°, 129.8°, 135°, In a tangential section (Fig. 7), the HSB are highly 140.2°, 146.3°, 153.4°, 161.6°, and 170.5°. uniform in width and sharply defined. The uniformity of width A tradeoff of the 24 azimuth algorithm is the greater is maintained through the occasional fission or termination of a tendency to average away small bends in the structure being band. The prisms of the darker bands are those of set B measured. This is a useful constraint here because the (descending from the EDJ ) and those of the lighter bands are objective is to document differences in the directions in which of set A (ascending). The transitional zone (containing prisms stresses acted in causing the cracks; fine irregularities in crack normal to the section) between adjacent HSB is narrow and paths are expected to be the result of material properties rather uniform in width (Fig. 8). The similarity in this view to the than stress directions. structure in rhinocerotoids (Figs. 9, 10; Boyde and Fortelius, A correction is applied to the raw frequencies to make all 1986, Fig. 10) is as strong as that noted for the transverse azimuths equally probable. In a digital image, a line connecting section. two diagonal pixels is longer than one connecting two pixels Horizontal decussation along the x or y axes. Without correction, unequal frequencies In the outer enamel, in which the prisms of set B of azimuths would be sampled along cracks of equal length generally run in an outward and slightly inclined direction running in different directions. toward the occlusal surface, a second zone of decussation exists. In polished radial sections, rather regularly spaced Astrapothere Molar Enamel bands can be seen within the outer enamel (Fig. 11). The long axes of these narrow bands run outward and slant somewhat The following descriptions are based on reflected light toward occlusal surface, in parallel with the general prism microscopic and SEM examinations of molar enamel sections direction of the outer enamel. At higher magnification (Fig. and surfaces in Astrapotherium from middle and earlier 12) the constituent prisms of these bands can be seen to

497 J. M. Rensberger & H. U. Pfretzschner

Figure 5. Astrapotheriwn s. /., UCMP 38825BL T, fractured,unground,un etched radial (vertical) sec tion of lower molar enamel. EDJ at right, occlusal to ward top (beveled left upper margin is wear facet). Vert ical decussation confined to approximately inner half of enamel thickness. Bar = Imm.

Figure 2. Astrapotherium sensu /ato, UCMP 38825B. This and subsequent UCMP specimens from La Venta Fauna, Colombia, S. A. (Friasian, late Miocene). This taxon is being described and assigned to a new genus by Steve Johnson and Rick Madden (Johnson, pers. comm.). Transverse polished and acid-etched straight section of lower molar enamel, viewed from cervical to occlusal. EDJ at bottom, outer surface at top. Bar = 200 µm.

Figure 3. Astrapotherium s. /., UCMP 38825B, orientation as in Fig. 2 but enlarged, showing well-defined boundaries of decussation zones in central region. A, zone with prisms running occlusally and outward; B, zone with prisms running cervically and outward (see Fig. 4). Prisms of zone B bend and become horizontal near upper part of micrograph while prisms of zone A continue in their occlusally directed course. Bar = JOO µm.

Figure 4. Astrapotherium s. /., UCMP 38825B, central region of Fig. 3 enlarged, showing reversed prism directions in zones A and B, horizontal attitude of prisms in transitional zones. Bar= 50 µm.

intercept the radial plane at a different angle than the prisms of the surrounding enamel. These prisms therefore have the same relationship to the vertical plane as horizontal HSB, that is, the plane of decussation is normal to the vertical plane. These zones of horizontal decussation are not to be confused with the faint continuation of vertical decussation that is sometimes present in the outer enamel of rhinocerotoids (Boyde and Fortelius, 1986). The angle of decussation (angle between crossing prisms) in the horizontal HSB (Fig. 12), is usually less than the angle of vertically decussating prisms in the inner enamel, and the zones occupy a smaller interval of the enamel thickness. The borders of the zones are usually less well-defined than those of the vertically decussating inner enamel, but in one specimen we examined the decussation is strong and the zones are well-defined and are separated by a thin zone of transitional prisms (Figs. 13, 14). The zones of horizontal HSB in the outer enamel represent a tissue distinct and unrelated to the vertically decussating zones of the inner enamel. An interval in which there is no decussation usually separates the two (Fig. 12). Previous studies found only vertical prism decussation in

498 Enamel Structure & Function in Astrapotheres

Figure 6. Astrapotherium s. /., UCMP 38825BLT, region of box in Fig. 5, frac ture

Figure 7. Astrapotherium s. /., UCMP 38063BM, tang ential ground section of low er molar enamel, viewed to ward EDJ, occlusal direction toward top. Outermost enam el removed to expose zones of vertically decussating prisms. Prisms in dark tracts cervically directed (zone B). Bar=200 µm.

Figure 8. Astrapotherium s. /., UCMP 38063BM, tang ential section, as in Fig. 7 but at higher magnification, showing horizontal prisms of transitional zone (normal to plane of section) bound ing occlusally and cervically running prisms of zones A and B. Bar= 100 µm.

Figure 9. Subhyracodon, UWBM 32019, a rhinocero tid, early Oligocene, N. A. Tangential section of ecto loph inner enamel of upper molar, viewed toward EDJ, occlusal toward top. Relief caused by variation in resist ance of enamel to air abras ion jet depending on prism direction. Bar= 200 µm.

Figure 10. Subhyracodon, UWBM 32019, tangential section of inner enamel of upper molar ectoloph, showing horizontal prisms of transitional zone and occlusally and cervically running prisms of zones A and B. Bar= 100 µm.

the cheek tooth enamel of rhinocerotoids (Rensberger and von Koenigswald, 1980; Fortelius, 1985; Boyde and Fortelius, 1986). However, horizontal decussation occurs in the outer enamel of rhinocerotoids; the condition there and its relationship to that in astrapotheres will be described in a separate study.

Astrapothere Canine Enamel

Canine tusk enlargement is characteristic of astrapotheres and goes back to the most primitive known genus, the early Eocene Trigonostylops (Cifelli, 1983b). The structure in the canine is distinct from that in the cheek teeth. The following descriptions are based mainly on SEM examination of sections

499 J. M. Rensberger & H. U. Pfretzschner

Figure 11. Astrapotherium s. I., UCMP 38825A Y, polished and etched Figure 13. Astrapotherium sp., GPIBO, Santacrucian (early Miocene, radial section of lower molar enamel showing vertical prism decussation Argentina), radial section through lingual margin of upper molar protocone in inner half of enamel thickness and banded pattern in outer enamel. in center of outer enamel, occlusal direction toward top, outer surface EDJ at left. Section occurs in region of strong horizontal curvature of toward left. Well-defined horizontal decussation with narrow transitional enamel plate. Bar= I mm. zones. Bar = 100 µm.

Figure 12. Astrapotherium s. /., UCMP 38825A Y, region of box in Figure 14. Astrapotherium sp., same specimen, section, and orientation Fig. 11, showing horizontally decussating prism zones in outer enamel. as Fig. 13, but to right of that frame, showing junction of inner and outer Bar= 100 µm. enamel. EDJ off frame at right. Cervically running prisms of zone B (at right) bend occlusally, begin decussating horizontally. Bar= 100 µm. because the complexity of the structure limits the interpretation Figure 15. Parastrapotherium, FMNH Pl5050LTI, Pico Truncado, of unsectioned specimens examined under reflected light. Santa Cruz, Argentina (Deseadan, early Oligocene), tangential section of Parastrapotherium lower canine enamel, direction of tip toward top, viewed toward EDJ. The enamel structure in the canines of the astrapotheres Showing aligned arcuate bands of differently directed prisms and sinuous examined is quite distinct from that in the molars. In tangential paths of cracks. Bands aligned longitudinally, wave crests aligned view of a lower canine of Parastrapotherium in which the outer transversely. Outermost enamel removed by air abrasion. Image contrast enamel has been removed (Fig. 15), sets of prisms running in results from differential "specimen collection" of electrons (Borchert, different directions form zig-zag bands, resembling a series of Vecchi, & Stein, 1991), here a blockage of secondary electrons produced waves. These bands run occlusally, i.e., longitudinally toward in depressions related to different prism directions (see Fig. 16). The the tip of the tooth (toward the top of the micrograph in this effect produces a map of prism divergence at magnifications too low to figure). The crests and troughs. in th_ese arcuate band_s are resolve prisms. Bar= 200 µm. aligned obliquely transverse (slantmg slightly dow!lward m the micrograph, left to right). This structure appears m tangenual Figure 16. Parastrapotherium, FMNH Pl5050LTI, higher magnification view to represent vertical HSB because the bands tend to be near frame of Fig. 15, showing wavelike changes in prism direction that continuous in the occlusal direction (Fig. 16), but the produce bands in Fig. 15. Bar= 50 µm. transverse alignment of the sinusoidal crests and troughs gives the appearance of horizontal HSB. The angle of decussat~on of Figure 17. Parastrapotherium, FMNH Pl5050LTI, region of box in Fig. the prisms in adjacent bands in the tangential section of Figures 15, showing relationship of prism directions in crack walls to crack 15 and 16 is often about 45°, but may be as much as 80°, as can directions. Along crack, local change in prism direction is associated with be seen along the margin of the crack in Figure 17. positions of knoblike bends in crack wall. Bar= 50 µm. The directions of the prisms of the inner part of the canine enamel in radial section (Fig. 18) resemble those in the Figure 18. Parastrapotherium, FMNH Pl5050LT3, polished and etched vertically decussating part of the molar enamel. The prisms run radial section of inner part of lower canine enamel, EDJ at left; showing more or less within the vertical plane in two directions that aspects of both horizontal and vertical decussation. Inner enamel with differ by 70° to 80°. One set slants cervically from the EDJ and prisms dominantly parallel to vertical plane of section, some running the other occlusally. outward and cervically (zone B prisms), others running outward and In transverse section (Fig. 19), the structure near the occlusally, as in vertical decussation. However, some prisms of inner EDJ (right side of the micrograph) contains transversely enamel turning horizontally to intercept plane of section at high angle elongated zones of prisms intercepting the plane of the section (e.g., near bottom of frame), as in horizontal decussation. at a high angle, separated by a zone of prisms running parallel Bar= 100 µm. with the plane of the section. The prism direction in each adjacent zone of high angle prisms is the reverse of the

500 Enamel Structure & Function in Astrapotheres

501 J. M. Rensberger & H. U. Pfretzschner

502 Enamel Structure & Function in Astrapotheres

Figure 19. Parastrapotherium, FMNH Pl5050LT3, polished and etched prisms in some areas run cervical to occlusal (e.g., above transverse section of lower canine enamel, from EDJ at right to outer center of figure) but in much of the fragment they tend to be surface at left; vertical decussation in inner enamel similar to that in more transversely aligned. Parallel sinuous bends, as seen molars, but zones bending within transverse plane and obliquely transverse below the center of the fragment, are similar to the pattern in in direction. In outer enamel zones become transverse. In central region Parastrapotherium, but overall the structure in tangential section there is some divergence of prism direction within horizontal plane, that lacks the regularity of the Oligocene form. is, some degree of horizontal decussation, but unlike that in molars. In transverse section (Fig. 21), it can be seen that the Decussation extends through greater part of enamel thickness than in prisms form a wavy pattern in which the prisms of adjacent molars. Bar= 100 µm. bands tend to parallel one another rather than decussate. The bending occurs mainly within the transverse plane (Fig. 22), Figure 20. Astrapotherium s. /., UCMP 29230AC, tangential section of although there is some vertical bending. The axes of the waves polished and etched upper canine enamel fragment, occlusal direction themselves bend and may be aligned in different directions toward right (arrow). Outer enamel removed. Gentle variation in direction (e.g., the slanting alignment of the wave left of center in the of prisms emerging on etched surface causes apparent "ridges" whose upper part of Figure 21). pattern reflects the differences in prism direction. Bar= 1 mm. In ground radial section (Fig. 23), the prisms in the La Venta form appear to form parallel waves whose axes trend Figure 21. Astrapotherium s. /., UCMP 29230AO, air abraded, etched obliquely in the cervical direction from the EDJ. In a ground transverse section of upper canine enamel, viewed occlusally, showing section it is difficult to distinguish between simple bending of wavelike bending of prisms that diminishes in outer enamel. EDJ at left. prisms and decussation. In a fractured, unground radial Bar= 100 µm. section (Fig. 24), oblique ridges represent the same cervically trending waves seen in Figure 23. In Figure 25 the shadowed Figure 22. Astrapotherium s. /., UCMP 29230AO, enlargement of central upper third of the image represents the same radial (vertical) region of Fig. 21. Bar= 100 µm. fracture surface shown in Figure 24; the lower two-thirds of the image is a transverse (horizontal) surface adjoining the Figure 23. Astrapotherium s. /., UCMP 29230AO, air abraded, etched radial surface. The undulating border (outer end marked by radial section of upper canine enamel, occlusal direction toward top, EDJ arrow) between these horizontal and vertical surfaces is at left. Wavelike bending of prisms, axes of bends running cervically and paralleled by prisms near the edge bending in the horizontal outward (from upper left to lower right). Because some prism bending plane. These undulations form the prominent ridges on the occurs in horizontal planes, ground section transects prism axes along vertical surface, indicating that the enamel structure consists certain sides of folds. Bar= 100 µm. mainly of bending prisms, not decussation. Von Koenigswald (1988:150), reported transverse HSB in an upper canine (astrapothere, gen. & sp. indeterminate). direction in the neighboring high angle zone, with the prisms of The structure he saw under light microscopy (pers. comm.) one slanting toward the EDJ and those of the other slanting may be similar to that in the La Venta form. Under light the toward the outer surface. The structure resembles that in the structure in the La Venta form appears to represent horizontal molar enamel because in both cases vertically decussating sets HSB. of high angle prisms are separated by zones of prisms parallel with the transverse section. Functional Interpretation The structure of the canine enamel in transverse section differs from that in the molars in several ways, however. The Molar Enamel HSB in the canines extend obliquely from the EDJ but bend In theory, horizontal HSB optimally resist the and become perpendicular to the latter toward the middle of the propagation of vertical cracks; in the primitive ungulates that enamel thickness, whereas in the molars the zones are have rounded cusps and dominantly vertical chewing perpendicular to the EDJ throughout their extent. The prisms movements (e.g. Arctocyon and other condylarths, and that rise (run occlusally) from the EDJ in the canine become Hyracotherium), the maximum stresses are expected to be horizontal near the middle of the enamel, whereas those that horizontal and tensile. This has been demonstrated by both run cervically from the EDJ flatten their trajectory somewhat mathematical and physical models (Fig. 26; Pfretzschner, but continue toward the outer surface in a slightly cervical 1988; Rensberger, in press). However for later, large direction. ungulates, finite element modelling demonstrates several ways In summary, the lower canine enamel of in which stresses may change (Pfretzschner, in press a, b; Parastrapotherium, like the molar enamel in all of the Rensberger, in press). In rhinocerotoid cheek teeth, the astrapotheres we have seen, is dominated by a pattern of direction of maximum stress is vertical, judging from the vertical decussation. However, as seen in transverse section, directions of premortem cracks as well as results of finite the HSB of the inner enamel are aligned oblique to the EDJ and element modelling of the principal stresses. The explanation are curved, rather than perpendicular and straight. In for this 90° rotation of the principal stresses is that the occlusal tangential section, the HSB have more or less regular wavelike structures have become thin lophs and the chewing force has bends whose crests are aligned in the vertical (occlusal) and become dominantly perpendicular to the plane of the loph, horizontal (transverse) directions so that they appear to causing the structure to bend along a horizontal axis (Fig. 26). combine aspects of vertical and horizontal decussation. The planes of decussation of the HSB, which are more or less La Venta Astrapothere horizontal in primitive ungulates, have also rotated 90° in Although upper canine enamel from Parastrapotherium rhinocerotoids, and this allows the enamel to maximally resist was not seen in this study, we examined enamel in the upper cracking under the new stress regime. canine of a late Miocene form, Astrapotherium s. l. from the The molars of astrapotheres, especially the uppers, La Venta fauna of Colombia. strongly resemble those of rhinocerotoids (Fig. 27). In both In tangential section, the La Venta astrapothere lacks the groups the labial part of the upper molar forms a thin, vertical, regular wavelike bending of differently oriented groups of bladelike ectoloph. Although the ectoloph is blunted by wear prisms present in Parastrapotherium. The different prism in the illustrated individuals, the plane of wear slopes linguad directions beneath the featureless region of the outer enamel and actually maintains an acute labial edge. In both groups the have an overall pattern that is much more random (Fig. 20) labial surface of the ectoloph is almost flat, marked only by a than that in Parastrapotherium. The bands of similarly directed narrow, vertical paracone fold and by a slight convexity near

503 J. M. Rensberger & H. U. Pfretzschner

Astrapotheria Rhinocerotoidea 27 paracone I rnetacone paracone I I metacone I

m:!astyle

labial

anterio,__t

\ protocone Astraponotus holdichi Hyracodon nebraskensis 1 L M (after Simpson, 1967) L M 1 (UWBM 27237)

domelike cusp model, loaded vertically

/ direction of chewing load _, - direction of max. stress cracks

504 Enamel Structure & Function in Astrapotheres

Figure 24. Astrapotherium s. /., UCMP 29230AO, upper canine enamel, ancestor that had these structural conditions, a similar set of unground, unetched radial fracture with rugose surface (surface of Fig. 23 selective factors must have been responsible. Strains resulting before air abrading). Occlusal direction toward bouom, EDJ at right edge. from vertical tensile stresses must have been dominant during Arrow indicates outer comer marked in Fig. 25. Bar= 100 µm. mastication, and occasionally these would be great enough to cause horizontal fracturing of the occlusal enamel and Figure 25. Astrapotherium s. /., UCMP 23230AO, upper canine enamel, consequent loss of chewing effectiveness in that locus. unground, unetched transverse fracture surface, top surface of enamel Vertical HSB seem to have provided an effective reinforcement segment in Fig. 24 looking occlusally, with vertical face of Fig. 24 against those strains. However, in the larger sample of visible as shadowed and ridged face at top of this Figure. Arrow identifies rhinocerotoid teeth examined, several segments of horizontally outer comer of edge between transverse and radial sections marked in Fig. fractured and subsequently worn enamel edges were found 24. EDJ at left edge. Bar= 100 µm. (Rensberger, in press; Figs. 12, 13), indicating that sometimes chewing induced vertical tensile stresses exceeded the strength Figure 26. Effect of extreme differences in shape of dental structure and of the material. direction of load on maximum stresses in enamel. Model on left depicts Canine enamel state in shell of stiff material covering less stiff interior, subjected to Premortem cracks are abundant near the tip of the lower dominantly vertical load (as in early Cenozoic ungulates). Models at right canine of Parastrapotherium (Figs. 31, 32). In this area the demonstrate deformation and stresses in narrow loph subjected to oblique enamel has a smoothly worn surface formed by abrasive load (as in rhinocerotoids and astrapotheres); maximum stress, which is contact with food or other environmental materials. Only a few tensile, is rotated 90°. of the cracks exposed at the surface do not have edges rounded by such abrasion. Figure 27. Occlusal views of upper molar in astrapothere and The frequency distribution of the premortem crack rhinocerotoid. Shaded areas represent dentin exposed by wear. directions on the labial enamel plate (Fig. 33) has prominent modes at 146.3° to 153.4° and at 71.6°, indicating at least two Figure 28. Astrapotherium s. I. upper molar, UCMP 38063, labial view. rather different sets of crack directions and stress regimes. Light micrograph showing light reflecting from planes of horizontal After calculating the premortem crack frequencies for the premortem cracks (arrows) in ectoloph enamel, near occlusal edge (above lingual enamel of the same tooth, it became apparent that the arrows). Light colored oblique line crossing crack at right is a surface dominant crack directions are similar to those near the dominant blemish. Vertical HSB visible near edge of enamel. Dark vertical cracks modal value (146°) in the labial enamel, but the distribution are postmortem. Bar = 1 mm. lacks the 71.6° mode exhibited by the labial cracks. Other cracks exist in the lingual enamel that initially had been Figure 29. Astrapotherium s. /., UCMP 38825B. Transverse (fractured, excluded because they do not reach the surface and could not unground, unetched) section of lower molar enamel looking occlusally, be differentiated on the basis of wear from cracks formed by with extremely rugose fracture surface through inner enamel. EDJ at postmortem events. When the azimuths of the undifferentiated bottom of figure. Bar = 1 mm. lingual cracks are added to those of the premortem cracks the 146.3° mode is not appreciably affected (Fig. 34), but the Figure 30. Astrapotherium s. /., UCMP 38825B, subregion of inner resemblance to the labial crack directions is strengthened by the enamel of Fig. 29 at higher magnification, showing fracture surface addition of azimuths near the missing 71.6° mode. It therefore plunging parallel to decussating prism directions in planes almost appears likely that the deeper cracks in the lingual enamel were perpendicular to one another. Bar= 50 µm. also formed by premortem events and the crack systems on the two sides are the result of transverse and obliquely longitudinal stress regimes. The cracks on the lingual side tend to bend the metacone (see also Fig. 1). ln both groups, the interior more strongly; the transverse cracks slant in a more cusps of the upper molar (protocone and hypocone) are longitudinal direction toward the bottom of Figure 35B, incorporated into narrow, posterolingually slanting lophs. producing a 123.7° mode adjacent to the 146.3° mode; bending The premortem cracks present in the astrapothere of the longitudinal cracks spreads their range of direction and ectoloph (Fig. 28) are horizontally elongate and shallowly flattens that mode. If the longitudinal cracks on the lingual side concave toward the occlusal edge of the enamel. These cracks (those from 39.8° to 71.6°) were due to bending from a labial have precisely the shape and orientation of the lenticular load, the deeper part of the enamel on the lingual side would premortem cracks that characterize the ectoloph of have greater tensile stress than the outer part, resulting in deep rhinocerotoids and significantly differ in these characteristics cracks like those observed that do not reach the surface. from cracks in the less derived cheek teeth of Mesohippus and The plane of the cracks in both labial and lingual enamel probably other perissodactyls (Rensberger, in press: Fig. 12). that reach the outer surface tend to dip away from the tip of the A transverse fracture through molar enamel in canine. The cracks of the major modal direction (146.3°) on Astrapotherium produces an extremely rugose surface across the labial enamel plate are reproducible by a load (LI) near the the inner enamel where vertical decussation occurs (Fig. 29). tip of the tooth, normal to the surface, and above the At this magnification the surface seems to consist of ridges and longitudinal axis (Fig. 35A). The enamel is platelike on either valleys, but at a higher magnification (Fig. 30) the surface is side of the canine (Fig. 35C). A finite element model of a seen to have stepped features in which the fractures parallel simple plate (Fig. 36) illustrates the relationships of such a load groups of prisms into the enamel before breaking across the and the resulting major stress, which is tensile and oblique to prism axes. The prisms and prism clusters act as fibers and the longitudinal axis, and dips toward the position where the much of the crack energy is absorbed by the pull-out of these load is applied; a resulting crack would be normal to the fibers. For fractures in this plane the enamel gains its direction of the maximum tensile stress and dip away from the toughness through a mechanism similar to that of the fabric in loaded end (Fig. 36, vertical section). Similarly, the cracks the thrust deflector used on the French Mirage jets, in which a near the 71.6° mode on the labial side and related cracks on the high-strength brittle fiber traces the strain lines in a brittle lingual side are reproducible by loads L2 in varying positions matrix (Besmann, et al., 1991). along the lower margin of the tooth and near the tip (Fig. 35A, The similar microstructures, stresses, and overall B). Because the transverse cracks intercepting the surface on occlusal shapes are all derived, not primitive conditions in both sides of the canine usually dip away from the tip, the astrapotheres and rhinocerotoids. Because these taxa were loading directions were probably reversed on opposite sides, isolated throughout their existences and do not share an with the labial loads exerted in a medial direction and the J. M. Rensberger & H. U. Pfretzschner lingual loads exerted laterally. Such loads would have been produced by pressing sometimes laterally and sometimes medially with the tips of the canines.

Conclusions

The enamel microstructure of the cheek teeth in astrapotheres is characterized by: •vertical decussation in the inner part of the enamel •linear boundaries of vertically decussating zones •uniformly defined transitional zone between the vertically decussating zones •horizontally decussating zones in the inner part of the outer enamel •horizontal decussation angle smaller than the vertical decussation angle The condition in astrapotheres is not distinguishable in any of these characteristics except the horizontal decussation from conditions that have been described for rhinocerotoids. The nature of the horizontal decussation in the outer enamel of rhinocerotoids, which is being studied, may make the resemblance between these groups even stronger. The gross structures of the molars of astrapotheres, especially in the upper dentition, are remarkably similar to those of rhinocerotoids in: •flatness of the ectoloph •direction of the ectoloph •shape of the ectoloph, including positions and shapes of the paracone, metacone, and metastyle •attitude of the occlusal facet on the ectoloph •chewing direction on the ectoloph •shape and direction of cracks in the ectoloph enamel The occurrence of an almost identical microstructural configuration in these two groups, an identity that cannot have occurred through a shared ancestry, is remarkable because this combination of structures has not been seen in any other group Frequencies of Crack Directions of mammals. The low probability that such a complex % 33 microstructure could have arisen independently in the two 25 groups by chance points to a similarity of selective events number of azimuths 1537 during the evolution of these masticatory mechanisms. 20 canine One of these events was a flattening and joining of cusps axis 146.3-153.4° to form narrow lophs. This event also occurred in other ungulate lineages, but in none of those that survived into the 1015 71.6 • later Tertiary was there development of vertical decussation L such as characterizes the rhinocerotoid and astrapothere molariform teeth. Analyses of the stresses and microstructural patterns in the other extinct taxa having vertical decussation ~ -L~-~..--+-,--,-, --.-, -r-1~"-r-r-r.---r-.-r--"r,..-:;=z---=-.-r=-r=, ,-,---,-,-,-, (e.g., pyrotheres and deperetellids) and in ungulates lacking 26.6 45 63.4 90 116.6 135 153.4 vertical decussation are needed. Parastrapotherium The wavelike bending of HSB in the lower canine enamel of Parastrapotherium and the bimodal crack directions lower canine, labial side and dipping crack planes suggest that the canines were subjected to different loading conditions, lateral and medial Frequencies of Crack Directions loading near the tip and loading at the anteroventral margin a % 34 short distance back from the tip. 25 The differences in structure of the canine and molar number of azimuths 1746 I enamel in these taxa appears to be a consequence of the 20 ~range for off-axis load - I differences in stress patterns. The distinction in the enamel 146.3° structure of the upper canine of Astrapotherium s. I. and the 15 lower canine of Parastrapotherium may be related in part to canine 10 123.7° differences in form which, even under similar loads, would axis result in differences in stress patterns. The absence of 5 decussation in the canine of Astrapotherium s. I. could be an 50.2° evolutionary loss resulting from the diminished importance of 0 J_r-~~~~~~~~~~~ crack inhibition in this larger form. The analogy to loss of 26.6 45 63.4 90 116.6 135 153.4 enamel in the tusks of late Cenozoic proboscideans comes to mind. Resolution of this issue will require more information Parastrapotherium about the differences in structure and crack patterns of the lower canine lingual side (reversed) upper and lower canines in each taxon.

506 Enamel Structure & Function in Astrapotheres

Parastrapotherium Lower Canine 35

cracks on lingual side near tip (reversed) cracks on labial side near tip C view looking cervically enamel labial at top enamel

layer of - layer of cementum cementum on - on surface cementum surface worn away worn away enamel enamel

Figure 31. Parastrapotherium, FMNH Pl 5050, labial view of right lower broken line = (off-axis load 36 canine near tip, showing cracks in enamel. Worn tip at right. Edge at top solid line= predicted normal to enamel direction of crack is margin of wear surface seen in lower part of Fig. 32. Light micrograph. direction of surface) L 1 Bar= 5 mm. maximum \ ""-

Figure 32. Parastrapotherium, FMNH Pl SOSO,lingual view of canine in stress ------T ~-~ ~ \ --\--),::-"::". /· "- ;-,-+-----,+--+-+-;-+--'-+-~~+-H-.....c.,+-_,_,+------ll Fig. 31. Worn tip at right. Dentinal surface at bottom worn by ": --. .-',, _..., ~ .-\,, _;,,__,,,· .. __/ "f. tooth-tooth contact, but dentinal surface at top and enamel surface in this end general with extensive wear caused by contact with food or environment. anchored --·. Light micrograph. Bar= 5 mm. ' ' -, Figure 33. Percentage frequencies of premortem crack directions in labial enamel from lower canine of Parastrapotherium, FMNH Pl5050. maximum stresses Fractional directions inherent in angles measured from digitized image (see are tensile thoughout Methods section for full list of directions). / I ----- enamel Figure 34. Percentage frequencies of crack directions in lingual enamel ----+--- from lower canine of Parastrapotherium, FMNH Pl SOSO. Most of cracks dentin recorded are identified as premortem by dental polishing. To those have been added deep cracks that cannot be objectively identified as such, but seem to be premortem, judging from similarity of direction to premortem cracks on labial side. Azimuths near 50.2° mode derived almost entirely from deep cracks. Acknowledgements

Figure 35. Crack patterns on labial (A) and lingual (B) enamel plates of We are grateful to the following individuals who lower canine of Parastrapotherium as viewed from labial side (pattern for provided access to specimens in their care: WA Clemens, JH lingual side flipped vertically); C cross-sections through canine, showing Hutchison and DE Savage (UCMP); JJ Flynn and W Turnbull confinement of enamel to labial and lingual plates. 0°, 90°, 180° (FMNH). S Johnson (UCMP) provided information derived correspond to azimuth directions in Figs. 33, 34. Ll indicates position of from his recent study of astrapothere systematics. The load normal to plates accounting for major modal direction of cracks. Load manuscript benefitted considerably from suggestions by AW at L2 (normal to plates at positions along lower margin of tusk) would Crompton and D Stem, Harvard University; WA Clemens, account for cracks of other modal direction. UCMP; M. Fortelius, University of Helsinki; and MC Maas, Duke University. JMR is grateful to Ray Olsen and Gene Figure 36. Directions of the major stress indicated by short solid line Ledbury of the Boeing High Technology Center, Bellevue, within each element in 3-dimensional finite element plate anchored at left WA. Mr. Ledbury's expertise was invaluable in setting up the end, loaded normal to surface at L 1. Maximum stress is tensile. Predicted joint SEM facility for the Department of Geological Sciences crack directions indicated by broken lines. Sectional view shows direction and Burke Museum at the University of Washington. of major stress and dip of crack plane typical of elements in right half of Similarly, Gerry O'Loughlan, Mike Diperna, Dave Meyers, model. and Chris Kuy! of AMRA Y Inc., and Clark Houghton of Secondary Images, Winchester, Ohio, generously provided valuable advice and assistance on various occasions. This

507 J. M. Rensberger & H. U. Pfretzschner study was supported in part by a grant from the GSRF, Koenigswald W von, Rensberger JM & Pfretzschner University of Washington. HU. (1987). Changes in the tooth enamel of early Paleocene mammals allowing increased diet diversity. Nature References 328:150-152. Korvenkontio VA. ( 1934 ). Mikroskopische Besmann TM, Sheldon BW, Lowden RA, Stinton DP. Untersuchungen an Nagerincisiven unter Hinweis auf die ( 1991 ). Vapor-phase fabrication and properties of Schmelzstruktur der Backenzahne. Ann. Zoo!. Soc. continuous-filament ceramic composites. Science 253: Zool.-Bot. Fennicae Vanamo 2. 353 pp. 1104-1109. . . Lucas PW & Luke DA. (1984). Chewing it over: basic Borchert AM, Vecchio KS & Stein RD. (1991). The use pnnc1ples of food breakdown. In: Food Acquisition and of charging effects in Al/AI20 3 metal-matrix composites as a Processing in Primates. Chivers DJ, Wood BA, Bilsborough contrast mechanism in the SEM. Scanning 13:344-349. A (eds). Plenum Press, New York, pp. 283-301. Boyde, A. 1976. Enamel structure and cavity margins. Maas MC. ( 1986). Function and variation of enamel Operative Dentistry 1:13-28. prism decussation in ceboid primates. Abstr., Amer. Jour. Boyde A. 1978. Development of the structure of the Phys. Anthr. 69:233-234. enamel in the incisor teeth in the three classical subordinal Maas MC. (1991). Enamel structure and rnicrowear: an groups in the Rodentia. In: Development, Function and experimental study of the response of enamel to shearing force. Evolution of Teeth: P.M. Butler, K.A. Joysey, eds. pp. 43-58. Amer. Jour. Phys. Anthr. ~:31-49. Academic Press, London. Pfretzschner HU. (1988). Structural reinforcement and Boyde A & Fortelius M. (1986). Development, crack propagation in enamel. In: Teeth Revisited: Proceed. of structure and function of rhinoceros enamel. Zoo!. Jour. the Vllth Intern. Symp. on Dental Morphology, Paris 1986, Linn. Soc. 87:181-214. Russell, D.E., Santoro, J.-P. & Sigogneau-Russell, D. Eds., Carbaj~J E, Pascual, R, Pinedo R, Salfity JA & Vucetich Mem. Mus. natn. Hist. nat., Paris (serie C) 53:133-143. MG. (1977). Un nuevo mamffero de la Formaci6n Lumbrera Pfretzschner, HU. (In press a). Enamel rnicrostructure (Grupo Salta) de la Comarca de Carahuasi (Salta, Argentina). and hypsodonty in large mammals. In: Structure, Function and Edad y correlaciones. Pub!. Mus. municip. Cien. Nat. Mar de! Evolution of Teeth. Proc. VIIIth Intern. Symp. on Dental Plata "Lorenzo Scaglia" Z:148-163. Morphology, Jerusalem, 1989. Freund Pub!. House. Cifelli RL. (1983a). Eutherian tarsals from the late Prfetzschner HU. (In press b). Biomechanik und Paleocene of Brazil. Amer. Mus. Novitates 2761: 1-31. Schmelzmikrostruktur in den Backenzahnen von Cifelli RL. (1983b). The origin and-affinities of the Grosssaugem. Palaeontographica A. South American Condylarthra and early Tertiary Litopterna Rasmussen ST, Patchin, RE, Scott, DB & Heuer, AH. (Mammalia). Amer. Mus. Novitates 2772:1-49. 1976. Fracture properties of human enamel and dentin. Jour. Clegg WJ, Kendall K, Alford NM, Button TW & Dental Res. 55:154-164. Birchall JD. (1990). A simple way to make tough ceramics. Rensberger JM. (1987). Cracks in fossil enamels Nature 347:455-457. resulting from premortem vs. postmortem events. Scanning Cook, J & Gordon, JE. 1964. A mechanism for the Microscopy 1:631-645. control of crack propagation in all-brittle systems. Proc. Roy. Rensberger JM. (in press). Relationship of chewing Soc. Land. 282A:508-520. stress and enamel microstructure in rhinocerotoid cheek teeth. Crompton AW & Hiiemae K. ( 1969). How mammalian In: Structure, Function and Evolution of Teeth. Proceed. teeth work. Discovery J.:23-34. VIIIth Intern. Symp. on Dental Morphology, Jerusalem, 1989. Crompton AW & Sun AL. (1985). Cranial structure Freund Pub!. House. and relationships of the Liassic mammal Sinoconodon. Zoo!. Rensberger JM, Forsten A, & Fortelius M. (1984). Jour. Linn. Soc. 85:99-118. Function~! evolution of the cheek tooth pattern and chewing Currey J. 1979. Mechanical properties of bone with direcllon m Tertiary horses. Paleobiology 10:439-452. greatly differing function. J. Biomech. 12:313-319. Rensberger JM & Koenigswald W von. (1980). Currey J. 1984. The mechanical adaptations of bones. Functional and phylogenetic interpretation of enamel Princeton University Press. 294 pp. rnicrostructure in rhinoceroses. Paleobiology .{i:447-495. Fortelius M. ( 1985). Ungulate cheek teeth: Stern D, Crompton AW & Skobe Z. ( 1989). Enamel developmental, functional, and evolutionary interrelations. ultrastructure and masticatory function in molars of the Acta Zoo!. Fennica 180: 1-7 6. American opossum, Didelphis virginiana. Zoo!. Jour. Linnean Hopson JA & Crompton AW. (1969). Origin of mam Soc. 95:311-334. mals. Pp. 15-72. In: Evolutionary Biology, Dobzhansky T, Teaford MF. (1988). A review of dental rnicrowear and Hecht MK & Steere WC (eds). Appleton-Century-Crofts, diet in modern mammals. Scanning Microscopy _2:1149-1166. New York. 309 pp. Van Valkenburgh B. (1988). Incidence of tooth Janis CM & Fortelius M. (1988). On the means breakage among large, predatory mammals. Amer. Nat. wh~reby ~ammals achieve increased functional durability of ill:291-302. their denllllons, with special reference to limiting factors. Biol. Van Valkenburgh B & Ruff, CB. (1987). Canine tooth Rev. 63:197-230. strength and killing behavior in large carnivores. J. Zoo!., Kawai N. (1955). Comparative anatomy of the bands Lond. 212:379-397. of Schreger._ Okijimas Folia anatornica Japonica 27:115-131. Young WG, McGowan M & Daley TJ. (1987). Tooth Koemgswald W von. (1980). Schmelzstruktur und enamel structure in the Koala, Phascolarctos cinereus:-some Morphologie in den Molaren der Arvicolidae (Rodentia). Abh. functional interpretations. Scanning Microscopy l: 1925-1934. senckenberg. naturforsch. Ges. Frankfurt 539: 1-129. Koenigswald W von. (1988). Enamel modification in Discussion with Reviewers enlarged front teeth among mammals and the various possible reinforcements of the enamel. In: Teeth Revisited: Proceed. of M.C. Maas: Regarding the brittleness of enamel, isn't the the Vllth Intern. Symp. on Dental Morphology, Paris 1986, composite nature of enamel a feature that functions to enhance Russell, D.E., Santoro, J.-P. & Sigogneau-Russell, D. Eds., its toughness and reduce its brittleness? In other words, there Mem. Mus. natn. Hist. nat., Paris, (serie C) 53:147-167. are general properties of enamel, other than its structural

508 Enamel Structure & Function in Astrapotheres arrangement, which make it a strong tissue. characteristics more primitive than those of the phenacodontids, Authors: Yes, the behavior of enamel as a composite lacking a hypocone, lacking at least vertical HSB and possibly material enhances its toughness, but structural arrangements HSB altogether. seem to be critical to the enhancements. Crystallites may behave as ceramic fibers in an organic matrix. Prisms behave M. Fortelius: As far as I know, the similarity of molar as ceramic fibers embedded in a matrix of another ceramic, the enamel structure between astrapotheres and rhinoceroses interprismatic enamel. Although prismatic fibers and extends to the level of prism pattern (both have Boyde's Pattern interprismatic enamel are both brittle materials, manufactured 3). Do you think that this is significant? To what extent would fibrous composites of two brittle materials withstand greater you consider the organization of apatite crystallites (as distinct deformation before fracturing than either alone. In fibrous from "prisms") to be important? composites in general, increased toughness seems to be gained Authors: The differences in crystallite organization may through the increased stress required to overcome the bonds define some broad functional differences among ungulates, and friction between the matrix and fibers running especially with regard to hypsodont forms, in which the perpendicular to the crack plane during fiber pullout (Currey, interprismatic enamel forms specialized sheets (Pfretzschner, in 1984: Fig. 2.6), in the delamination of the fibers and matrix at press a,b). Interprismatic enamel is meagerly developed in the crack tip, and in overcoming the reduction in stress astrapotheres and rhinocerotoids. It is possible that an early concentration because of blunting or branching at the crack tip difference in crystallite organization influenced the direction (Cook & Gordon, 1964). This applies to simple prismatic evolution took during later evolutionary events, but more data enamel. HSB cause groups of prisms to behave like fibers, as is needed to test this hypothesis. indicated by the roughness of the fracture surface produced by pullout of clusters of prisms across entire HSB (Fig. 30) and W.A. Clemens: The upper and lower canines of by the deflecting or branching of cracks by HSB (Pfretzschner, astrapotheres differ in morphology and angle of implantation. 1988: Fig. 13). In all of the above cases, the structural Can you determine if the differences between the enamel organization of the enamel components with respect to the microstructure of the lower canine enamel of Parastrapotherium stresses seems critical to the gain in toughness. In spite of and the upper canine enamel of Astrapotherium reflect these gains, brittleness still seems to be a major factor limiting differences in position in the dentition of the teeth compared or the strength of enamel. Even enamel with HSB exhibits about might differentiate canine enamel of the species? the lowest work to fracture among skeletal materials: 200 J/m 2 M.C. Maas: What are the relative contributions of in human enamel compared to 270 for human dentin, 1710 for biomechanics and phylogeny to differences in the enamel bovid limb bone, and 6186 for cervid antler (Rasmussen, et structure of Astrapotherium upper canines and al., 1976; Currey, 1979). Parastrapotherium lower canines? Authors: Until the structures in opposing canines in both taxa M. Fortelius: From the point of view of evolutionary theory are studied, we can only note that the structure in the lower and systematics it might be useful to summarize the anatomical canine of the earlier of the two taxa, Parastrapotherium, differences between astrapotheres and rhinos. Rodents and combines an aspect of the structure in the upper canine of primates came into South America from the outside, and for the Astrapotherium (wavelike bending of prisms) with that in the Paleocene the xenungulate-dinocerate relationship lurks in the molars of both taxa (vertical decussation). To understand the background. Briefly comment on why astrapotheres are not phylogenetic influence, we not only need to know the structure simply displaced rhinos. in the opposing canines of these taxa, but also the structure in W.A. Clemens: What group included the last common other, preferably more primitive, astrapotheres. The differences ancestor of astrapotheres and rhinocerotoids? What evidence in morphology of the upper and lower canines examined, is available demonstrating that this last common ancestor lacked which extends also to a difference in cross-sectional outline, the specialized HSB of the astrapotheres and rhinocerotoids the upper being subcylindrical, the lower having platelike studied? enamel, makes the mechanics different. Authors: These questions center on the issue whether astrapotheres and rhinocerotoids might not share a common M.C. Maas: Are the differences that you identify between ancestry that already exhibited some specialization of the HSB astrapothere molars and Parastrapotherium canines explainable shared by later members of these groups. Perhaps the most only in terms of mechanical differences in the function of the conspicuous feature indicating that astrapotheres did not derive teeth, as you suggest, or is there a phylogenetic influence? from rhinocerotoids is the absence of a distinct hypocone in Authors: The main difference is that the HSB in the canine of primitive members. In primitive astrapotheres (Fig. 27 ), Parastrapotherium bend whereas those in the molars of unlike primitive rhinocerotoids, the hypocone is poorly astrapotheres are straight. Both have vertical decussation. The developed. In the Trigonostylopidae, which have traditionally condition in primitive ungulates seems to be straight HSB and been included in the Astrapotheria, although placed in a distinct horizontal decussation. The condition in the Parastrapotherium order by Simpson (1967), hypocone and metaloph are virtually canine therefore differs from the presumed primitive condition absent (instead of a hypocone, there is only a slight expansion both in bending of the decussation planes and in the attitude of of the posterolingual cingulum). Recent studies (Carbajal et these planes. The more complex HSB alignment in the canine, al., 1977; Cifelli, 1983a) support a close relationship between which gives it the ability to resist more complex stress the Astrapotheria and the Trigonostylopidae. Cifelli (1983b) directions, is predicted by the complex stress regime in the regarded Trigonostylops as the most primitive astrapothere and canines in contrast to the uniform chewing direction and stress noted that the structural differences cited by Simpson are alignment in the molars. It would be genetically more mainly primitive characteristics. The absence of a hypocone in economical if the canines and molars had the same enamel Trigonostylops makes remote the likelihood of a relationship to microstructure, so the available evidence points to a mechanical the most primitive members of the Rhinocerotoidea (the influence. This relationship can be further tested: should it turn Hyrachyidae), which retained the strong hypocone and out that the microstructural condition in a primitive astrapothere adjoining metaloph of primitive perissodactyls and in which the canine is still small and circular in cross-section phenacodontid condylarths. Hyracotherium lacks vertical HSB and was loaded only at the tip has aspects of the structure in (Rensberger & Koenigswald, 1980). The latest common Parastrapotherium, that would be evidence of a phylogenetic ancestor of astrapotheres and rhinocerotoids therefore appears influence dominating the mechanical relationships. to have been an early Paleocene condylarth with dental J. M. Rensberger & H. U. Pfretzschner

M.C. Maas. Can the authors quantify the percentage of vertical decussation and horizontal decussation? Was there any variation within a tooth? If so, could the variation be related consistently to a certain position on the tooth? Authors. There may be variation in the development of horizontal decussation from one region of a tooth to another, but we do not have enough observational data at this time to determine whether this is true. Because of the limited material and the expectation of variation from region to region we did not feel that quantification of decussation patterns would be useful information at this time. However, we hope to pursue these problems when more data can be assembled. In the meantime, we can only note that all the material we have seen under both light and electron microscopy has had the horizontal decussation, when present, confined to an interval within the outer enamel, as shown in several of the micro graphs.

Additional Reference

Simpson, GG. 1967. The beginning of the age of mammals in South America. Part 2. Bull. Amer. Mus. Nat. Hist. fil:1-260.

510